Nick Maciel

The difference lies in defining the rules and inputs the tools work from—and assigning someone to own that process.

• Most companies use AI, the posts argue, but far fewer scale its value beyond adoption.
• The practical conclusion: the job shifts from producing the work to defining the system that produces it.

In the Brand Profiles post, Nick Maciel contrasts the old Performance Max process—load assets and hope the algorithm performs without making them look off-brand—with a new one: define the brand once, and AI produces the variants.

• The closing line makes the shift explicit: the job moves from writing twenty versions of an ad to defining the system that writes them.
• The agency post makes the same argument in terms of capacity: teams spend about one full-time-equivalent week a month on reports clients skim in ten seconds, while agencies that automate reporting recover roughly 40 hours a month.

Read original · Original 2

AI assembles what it is fed, the posts argue, so poor inputs produce poor output at greater volume.

• Feed hygiene, naming and asset standards determine whether a new creative feature helps or adds clutter.
• Timing matters, too: accounts that are ready move first.

The Demand Gen post ends with a pre-rollout checklist: clean the product feed, standardise image sizes, tighten the naming structure and remove outdated promotional assets from the asset library.

• Its summary: garbage in means garbage out.
• The Brand Profiles post lists the equivalent training inputs: five to ten high-performing headlines, three to five strong descriptions and the best static images.
• The warning is the same: bad inputs produce bad outputs.

Read original · Original 2 · Original 3

When images and copy can be recombined automatically, the limit becomes how well those inputs are organised. Nick Maciel recommends thinking in blocks rather than finished ads, so the same images and messages can be mixed into audience-specific combinations without a new shoot or edit.

The Performance Max image post describes taking a hero product from three to five base photos to more than fifty assets in under an hour.

• The process: generate ten to twenty AI backgrounds, then use image expansion for vertical Shorts, landscape display and portrait placements.
• The Demand Gen post outlines the upload: three to five hero product images, three lifestyle shots, five to ten short benefit bullets, three offers and three calls to action.
• It also recommends directions such as leading with social proof or showing the product in use in the first three seconds.

Read original · Original 2

They also call for guardrails. Generated creative needs review for brand fit, visual errors and claims, and a person remains accountable for what goes live.

The Performance Max image post sets out a campaign experiment: current assets as the control, AI backgrounds plus expanded images as the variant.

• It measures conversion value per cost, ROAS and click-through rate by asset group.
• The post then describes checking which backgrounds earn Best ratings in the asset details view and promoting the winners into search companion images, other placements and social briefs.
• Guardrails cover off-brand colours or props, odd hands or faces, and false product claims.
• The Demand Gen post applies the same logic, comparing AI videos with static Demand Gen ads and existing manual video on view rate, click-through rate, cost per view and cost per action.

Read original · Original 2 · Original 3

Define fonts, colours and voice once, and the system has a spine to generate against; testing aggressively and staying on brand no longer pull in opposite directions. Someone still has to own the profile and review output on a schedule.

The Brand Profiles post explains how to find the feature under Performance Max, then Settings, then Brand, and add a logo, colours, fonts and brand voice.

• It defines voice through tone, example copy, phrases to avoid and phrases to prefer.
• Its four-week rollout runs from building the profile out of the existing style guide to enabling asset generation on one campaign, pausing off-brand assets, then expanding to more campaigns or countries.
• Weekly asset review, negative keyword lists, brand-safe placements and compliance checks remain on the required list.

Read original

The claim is less about selling more than finding parts of the route to market a business has never built. A companion thread frames manufacturing and product-development mistakes as learning opportunities rather than failures—the same review-and-correct instinct applied to operations.

In the Nuts Factory post, Nick Maciel describes visiting a business with strong retail and e-commerce operations and identifying the absence of a distribution arm as a limit on growth and market reach.

• The post concludes that what you do not know can hurt you and that regular strategy reviews matter, no matter how successful a business appears.
• A related post invites readers to share their product-development hurdles; another treats common manufacturing mistakes as a basis for improving quality in the year ahead.

The posts cover two subjects rather than one line of development.

• One reflects on entering 2024; the Cadversity posts read as a consulting-facing thread.
• The Google Ads posts discuss features described as working, rolling out or in testing, making them the more recent material.
• What connects the strands is an instinct to review the whole system, find the missing piece and fix the process rather than the symptom.

Read original · Original 2 · Original 3

A twelve-part breakdown of Google testing generative AI inside Demand Gen.

• The mechanic: feed product images, headlines, descriptions and existing feed assets, and Google's AI assembles short videos for YouTube Shorts, in-stream and Discover.
• Maciel identifies who benefits most, including DTC brands with strong product photos, local businesses on limited budgets, and B2B teams with content but no video crew.
• The post lists the inputs to prepare and suggests an asset structure: three to five hero product images, three lifestyle shots, five to ten short benefit bullets, three offers and three CTAs.
• It offers prompt phrasings, such as leading with social proof or showing the product in use in the first three seconds, and recommends measuring view rate, click-through rate, cost per view and cost per action against static Demand Gen ads.
• A pre-rollout checklist covers cleaning the product feed, standardising image sizes, tightening naming and removing outdated promos from assets.

For brands planning a campaign

The asset-block structure and metric list translate almost directly into a creative testing roadmap.

• The post shows how this specialist frames the problem: well-structured inputs matter more than production craft.
• Useful for briefing an AI creative test and assessing whether a candidate thinks in testable hypotheses rather than tool announcements.

A practical walkthrough of AI image editing in Performance Max: image expansion adapts one product photo for vertical Shorts, landscape display and portrait placements, while AI background generation creates seasonal backdrops, lifestyle setups and colour variations.

• Maciel recommends three to five base photos per hero product, ten to twenty generated backgrounds, and image expansion to cover formats—taking an account from five assets to fifty or more in under an hour.
• The test design is specific: run a PMax experiment with current assets as the control and AI backgrounds plus expanded images as the variant.
• Judge results by conversion value per cost, ROAS and click-through rate by asset group, then keep the winners and pause the rest.
• The post also covers checking asset ratings in the campaign's detail view and guarding against off-brand colours, odd hands or faces, and false product claims, with brand-safe prompt templates reused for each product line.

For brands planning a campaign

Direct input for an account audit: asset volume, image expansion coverage, experiment structure and guardrail review are all checkable. It also offers a strong interview question: how does Maciel distinguish AI-generated creative that lifts performance from creative that only adds volume?

This post covers Brand Profiles inside Performance Max, under Settings, then Brand.

• Maciel lists logo, colours, fonts and brand voice as inputs; voice is defined through tone, example copy, phrases to avoid and phrases to prefer, so generated copy reflects an existing style instead of defaulting to generic ad language.
• The older approach—loading assets and hoping the algorithm performs—is contrasted with defining the brand once so AI can produce variants.
• The week-by-week rollout plan is to build the profile from the style guide, enable asset generation on one campaign, pause off-brand assets and keep the winners, then expand to more campaigns or countries.
• Training inputs include five to ten strong headlines, three to five descriptions and the best static images.
• Guardrails include weekly asset review, negative keyword lists, brand-safe placements and compliance checks for regulated offers.
• The post argues that the job shifts from writing twenty ad versions to defining the system that writes them.

For brands planning a campaign

Raises the measurable question of brand consistency versus ROAS, as well as who owns brand profile updates. Useful interview material on how a paid media specialist handles brand control when creative generation is automated.

It explains why Maciel frames AI as a change in how work gets done, not just a tool.

A shorter post aimed at agency owners.

• Maciel argues that teams spend roughly one full-time-equivalent week each month building reports clients skim in ten seconds, and that time either goes back to the agency or to a competitor.
• Maciel cites figures that nine in ten US marketing agencies now use generative AI, about half use agentic AI for execution, and 74 percent of companies still struggle to scale value past adoption.
• Agencies that automate reporting recover an average of 40 hours a month, Maciel says, framing that time as capacity to absorb work larger shops are cutting rather than a reason to shrink a team.
• The post asks owners of ten-to-fifty-person shops where those hours would go.
• The figures have no named study, so treat them as Maciel's claims rather than cited findings.

For brands planning a campaign

Relevant for agency efficiency or reporting redesign, less so for ecommerce account audits. The agency-facing audience matters if a campaign needs agency buyers rather than direct brand marketers.

A post about a visit to the Nuts Factory for a business audit.

• Maciel identifies strong retail and e-commerce operations but no distribution arm, which they see as a limit on growth and market reach.
• Writing as CEO of Cadversity, Maciel concludes that regular strategy reviews matter for successful businesses.
• Companion posts make a similar move, inviting readers to share product-development and manufacturing challenges and framing manufacturing mistakes as learning opportunities.

For brands planning a campaign

Confirms a second, consultancy-facing strand beyond Google Ads, but the support is thin: no numbers, scope or outcome.

• For a paid media need, treat these posts as context, not proof of consulting depth.
• Ask for named case results or a service page before relying on this experience.

Each post states where the feature lives in the interface, what to prepare in an account before it rolls out, and which metrics to watch.

Numbered sequences move from what changed, to why it matters, to who benefits, to specific account actions.

For brands planning a campaign

This is the closest match to an audit-plus-roadmap engagement. The posts show working familiarity with the exact surfaces a client account would need reviewed — Demand Gen assets, PMax asset groups, and Brand Profile settings — and name the preparatory steps an audit would check first, such as cleaning product feeds, standardising image sizes, tightening naming structures, and removing outdated promotional assets.

Examples: Source 1 · Source 2 · Source 3

It describes checking which backgrounds earn Best ratings in the asset details view, promoting winners into search companion images and other placements, and reviewing AI output for off-brand colours or props, distorted hands or faces, and false product claims.

For brands planning a campaign

This can be lifted almost directly into a paid test plan: it defines a control-and-variant experiment, the asset volumes to produce, and the metrics to read — conversion value per cost, ROAS, and click-through rate by asset group.

Examples: Source 1 · Source 2

It names who benefits most (DTC brands with strong product photography, local businesses with limited budgets, B2B teams with content but no video team), the asset blocks to prepare (hero images, lifestyle shots, benefit bullets, offers, calls to action), prompt-style direction such as leading with social proof and showing the product in use in the first three seconds, and metrics to compare against static Demand Gen and manual video creative.

The post describes a feature Google was testing, not a documented general rollout. Confirm availability inside the account before making it part of a client plan.

For brands planning a campaign

Covers the creative-roadmap half of a brief. The post ties AI video output to a measurement set — view rate, click-through rate, cost per view, cost per action — and a comparison against existing creative, providing the basis for a defensible test rather than a production experiment.

Examples: Source 1

It frames those hours as capacity to take on work that larger shops are cutting loose, not as a reason to run a smaller team.

For brands planning a campaign

This is the clearest example of how Nick Maciel speaks to agency operators — the same audience the brief needs to reach. Useful for judging tone and framing a test plan around recovered capacity and fulfilment rather than tooling for its own sake.

Examples: Source 1

They cover a business audit of a retail and e-commerce business that identified a missing distribution arm as a limit on growth, an open invitation to share product-development hurdles, and a reflection on common manufacturing mistakes as learning opportunities for the year ahead.

These posts make broad claims, such as that a business lacks a distribution arm, without supporting detail. Ask for the underlying analysis before treating them as evidence of analytical depth.

For brands planning a campaign

A second subject area in the same feed.

• It shows an operations- and strategy-facing side that could matter if a brief involves manufacturing or product clients.
• For an ecommerce paid media brief, it offers context on how Cadversity is positioned; ask for a relevant technical case study if manufacturing expertise is central to the decision.

Examples: Source 1 · Source 2 · Source 3

They routinely name interface paths, asset quantities, and metric names, and invite a reply — for example, asking what a team would do with 40 recovered hours a month or whether an asset workflow could keep up with competitors shipping far more creative.

Numbered listicles with headline claims and supporting bullets, designed to be read and acted on in one sitting.

For brands planning a campaign

The predictable structure makes the posts easy to reuse as a briefing template. A review session can follow the same order: what is available, what to prepare, what to test, what to measure, and what to do this week.

Examples: Source 1 · Source 2 · Source 3 · Source 4

In one post, they ask readers to share their product-development hurdles and invite them to explore how Cadversity could be “the partner you need,” promoting the company’s services. Another post uses the Cadversity hashtag while discussing common manufacturing mistakes.

View example

The post frames the trip as advisory work but does not say whether Cadversity was engaged, what it delivered, or whether the relationship continued. Ask about the engagement’s scope and outcome.

View example

Both writers frame AI image generation as creative operations at scale rather than a novelty, and sell the idea through specific protocols and named workflows.

Mike Futia's material centres on one image-generation tool and its playbook prompts, with a guide offered in exchange for a comment, so the posts double as lead capture. Nick Maciel focuses on the ad platforms themselves — Demand Gen, Performance Max assets and Brand Profiles — and spends more time on account preparation, asset-block structure and which metrics to read than on any specific third-party tool.

Read original · Original 2 · Original 3 · Original 4

Brand handling in Performance Max is a shared subject, including whether to exclude brand from PMax and how to read brand versus non-brand clicks.

Martin Davies is strongest on auditing and diagnosis, including third-party scripts that show brand share of PMax traffic; a single tip often stands alone. Nick Maciel builds longer, numbered walkthroughs that cover what changed, who benefits and what to do this week, with more emphasis on producing and grading creative than auditing an underperforming account.

Read original · Original 2 · Original 3 · Original 4

Both writers treat PMax as something to configure deliberately rather than hand over to the algorithm, and explain the reasoning behind each step.

Robin Tesselaar's argument is financial: ROAS is not return, break-even ROAS depends on gross margin, and the decision metric is profit-based, using nCAC and POAS. Nick Maciel focuses on the creative inputs and brand system feeding Performance Max campaigns, not the margin maths or profit reporting around them.

Read original · Original 2 · Original 3 · Original 4

The posts offer practical playbooks for turning one product photo into dozens of campaign-ready images: uploading up to 14 reference images to keep logos, labels and packaging accurate; generating lifestyle scenes from a single SKU; swapping models to test different buyer personas; rendering readable text inside the image in any language; relighting a winning ad to fight fatigue; and adapting images to different aspect ratios without cropping the product.

• Nick Maciel's posts cover AI creative inside Performance Max and Demand Gen, but focus on directing the account.
• Futia's material gets into the generation workflow, with named protocols (Anchor Stack, Mannequin Swap, Hook Factory, Science Ad) and a sample output.
• For an ecommerce client whose Performance Max ceiling is asset supply rather than bidding, Futia is the more specific choice.

The posts focus on creative operations and lead generation, not account audits.

• Neither explains bidding, budget structure, measurement or profit analysis, so an audit would still require Nick Maciel, Martin Davies or Robin Tesselaar.
• Both posts link to the same downloadable playbook behind a comment request; the public text offers a sample, not the full method.
• No client outcomes appear in the posts.

Examples: Source 1 · Source 2

The posts answer two audit questions that Nick Maciel's posts leave open.

• First, how to tell whether a sales dip comes from lower market demand or from competitor activity: run an impression share report by day for the past 30 days or from before the dip, and if share holds steady the problem is demand, while a falling share points to competitive pressure or a competitor promotion.
• Second, whether brand traffic should be excluded from Performance Max, judged with a script that separates branded and non-branded clicks, conversions and share of brand revenue over time, and a recommendation to run brand through search campaigns instead if brand is taking a high share of Performance Max traffic.
• If the agency need is to understand why an account slowed down, or whether it is paying for conversions it would have received anyway, this information addresses the question directly.

The posts don't cover AI asset generation, image expansion, brand profiles or video automation, so they can't provide a creative testing roadmap on their own.

• One post largely points to a third-party script rather than describing an original method.
• With only two posts, the sample is narrow.
• Pair this work with Nick Maciel's creative playbooks, or with Mike Futia when creative volume is the constraint.

Examples: Source 1 · Source 2

The posts argue that ROAS is not return: it is revenue divided by ad spend, before cost of goods, shipping, returns and fulfilment.

• They give the arithmetic: at 35% margin, break-even ROAS is 2.86x, so anything below that loses money.
• Tesselaar recommends measuring profit on ad spend and new-customer acquisition cost.
• The posts also warn that in brand-dominant accounts, 60 to 80% of attributed conversions may have happened without the ad, and that average ROAS hides marginal returns.
• Worked numbers show a lower headline-ROAS channel generating more profit on incremental spend.
• The recommended Performance Max structure is specific: protect branded traffic with a dedicated exact match search campaign so Performance Max competes only for cold traffic; split asset groups by margin with separate target ROAS settings; feed CRM and lifetime-value audiences; use conversion value rules so a $20 sale and a $200 sale aren't treated as the same signal; and use search themes only for discovery terms not already captured.

The posts don't cover AI creative generation or asset testing, so they won't help on their own if the client's asset library is the constraint.

• They describe structure and metrics but provide no account-level results behind the specific figures.
• Use this work to set the measurement frame, then populate it with the creative tests described by Nick Maciel.

Examples: Source 1 · Source 2